Ervin Yohannes, Alfito Mulyono, Aditya Prapanca, Fitri Utaminingrum, Timothy K. Shih, Chih-Yang Lin, Kahlil Muchtar
Traffic is a key element in the transportation system. However, the increase in road accidents, which can be attributed to people's lack of knowledge about traffic, is a growing concern. The primary solution to address this problem is to enhance traffic knowledge. The application of artificial intelligence, particularly object detection methods using detector-based deep learning, has proven efficient in real-time object detection. In this research, object recognition is performed using YOLOv8 (You Only Look Once). These models are trained and tested for their performance in detecting traffic signs in Indonesia. The results show that the mAP 50 and mAP 5 0 - 9 5 values were 9 9. 5 0 % and 9 9. 0 1 %, respectively, for the YOLOv8 model. This demonstrates that YOLOv8 is the best-performing model for traffic sign detection in Indonesia. © 2024 IEEE.
State University of Surabaya, Informatic Engineering, Surabaya, Indonesia; University of Brawijaya, Informatic Engineering, Malang, Indonesia; National Central University, Computer Science and Information Engineering, Taoyuan, Taiwan; National Central University, Mechanical Engineering, Taoyuan, Taiwan; Syiah Kuala University, Electrical and Computer Engineering, Banda Aceh, Indonesia
